MiniMax M3.1 Flash comes to Tenstorrent Cloud with open weights
10/24/2026 — 10/24, 06:20·3 sources (all)·4 reports (all)
Key takeaways
- MiniMax has released M3.1 Flash, a 428B-parameter MoE model activating about 23B parameters per token, clocking over 200 tokens per second a…
- MiniMax has not disclosed performance, pricing, or availability details.
Background & analysis
MiniMax released MiniMax M3.1 Flash on October 19, 2026, describing it as a 428B-parameter MoE model that activates about 23B parameters per token, runs at over 200 tokens per second, and is priced at $0.15 per million input tokens — roughly 70% cheaper than its predecessor. Those figures come from engineering notes disclosed by the company. According to the X account TestingCatalog, MiniMax has published MiniMax M3.1 Flash weights on Hugging Face, noting that the model had previously been tested in stealth on OpenRouter under the codename Space Bunny.
MiniMax then used X to announce the rollout. First, Fireworks AI brought M3.1 Flash to serverless preview on day 0, positioning it for multi-step agent workflows. Later, MiniMax announced that MiniMax M3.1 Flash is now on Tenstorrent Cloud, where it runs on Tensix silicon and an Ethernet scale-out architecture, with open weights and an open-source stack. That is where the matter stands, as the Tenstorrent Cloud announcement did not add performance, pricing, or region details.
AI-generated from 4 reports · updated 51 minutes ago
Latest turnMiniMax says M3.1 Flash is now on Tenstorrent Cloud, running on Tensix silicon and Ethernet scale-out with open weights and an open-source stack. The announcement did not give performance, pricing, or availability details.
Related AI tools
- Heat index
- 243
- All sources
- 3
- All reports
- 4
- First seen
- 7 hours ago
Reports on this story headlines open the original
MiniMax says M3.1 Flash is now on Tenstorrent Cloud, running on Tensix silicon and Ethernet scale-out with open weights and an open-source stack. The announcement did not give performance, pricing, or availability details.
X：MiniMax (@MiniMax_AI)First-partyAI score 56MiniMax says Fireworks AI launched M3.1 Flash on serverless preview on day 0 for multi-step agent workflows, alongside a rebrand.
X：MiniMax (@MiniMax_AI)First-partyAI score 56MiniMax released M3.1 Flash on October 19, 2026, claiming a 428-billion-parameter MoE model that activates only about 23B parameters per token, runs at over 200 tokens per second, and costs $0.15 per million input tokens, roughly 70% cheaper than its predecessor. Those figures come from the vendor's engineering write-up.
DEV Community · AIAI score 66
According to TestingCatalog on X, MiniMax has published the model weights of MiniMax M3.1 Flash on Hugging Face. The model features 428B total parameters with 23B active per token, previously known as the stealth model 'Space Bunny' on OpenRouter.
X：TestingCatalog (@testingcatalog)AI score 63
Other stories people are talking about
- Heat index 82MiniMax M2.7 uses self-critique training to cut agent costs1 sources
- Heat index 709SurgeKimi K3.1 Pro launches with API access; weights due by Nov 48 sources
- Heat index 536NewSurgeOpenAI resumes frontier training on isolated offline network6 sources
- Heat index 301Major US banks tighten access for AI agents amid unauthorized activity5 sources
- Heat index 287DeepSeek closes around CNY 10B funding round led by CATL and Tencent4 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion